Abstract of Gunnar Schumann: Epistemische Rechtfertigung und Wahrheit als Empfehlung (Epistemic Justification and Truth as Commendation), Münster: Mentis 2013.
The core idea of my doctoral thesis is: epistemic justification and the truth predicate are (just
like ethical concepts) primarily concepts of commendation and commendation should be understood as an instance of imperative language use and not of assertoric language use. When I
say ‘primarily’ then I mean that evaluative notions also can and do have descriptive meaning,
for when we ask “Why do you evaluate X positively?” we expect as an answer some descriptive criteria of X in virtue of which X is evaluated positively. And, if standards of evaluation
are known and commonly spread, evaluative judgements also can be used to inform recipients
that X has some descriptive property. And, we make use of this interplay between prescriptive
and descriptive meaning of evaluative terms when we teach or learn standards of evaluation.
The relation in which the prescriptive and descriptive meanings of evaluational concepts stand
is that of supervenience. Likewise “justified” and “true” are evaluative properties which supervene on descriptive ones. As an analogue to Hare’s Universal Prescriptivism, I call this
“Epistemic Universal Prescriptivism”.
The old idea of epistemology as an “ethics of belief” is thus spelled out. But, to lay stress
on the primary non-descriptive meaning of such terms and their supervenient character is no
end in itself but has some very important consequences for the shape of substantial theories of
epistemic justification and theories of truth. Let’s first consider epistemic justification:
First, and that’s a very obvious consequence, epistemic justification cannot be defined
(solely) with the help of other evaluative terms, because definitions of this sort would not be
illuminating at all. We need theories of justification which deliver us with criteria of the application of justification.
Second, the supervenient character of justification may help to resolve the well known
epistemological struggle between Fundamentalists and Coherentists by delivering an intermediate position. Coherentists (if I may simplify a little bit) are wrong in holding that justification of a certain belief could be derived from other justified beliefs or a whole system of beliefs which is somehow already justified. Justification is a supervenient property which has to
“rest”, as it were, on solid descriptive criteria. This is so, because it is always reasonable to
ask “But why do you call that belief ‘justified’?” (as it would be reasonable to ask “But why
do you call that thing ‘good’?”) And then we expect the indication of some descriptive properties, so that it wouldn’t be satisfying at all to get the answer: “Oh, there is no further reason.
It just is justified/good” respectively “It stems from a justified/good origin.” Fundamentalists,
on the other side, are wrong in holding that these solid descriptive criteria deliver nothing but
the meaning of justification outright. Fundamentalism mainly comes in the form of Reliabilism (or “External Foundationalism”) which holds that the status of a belief to be justified
is semantically equivalent to being the result of a reliable cognitive process. The mistake of
the Fundamentalist’s position is that descriptive criteria for justification are not at all semantically equivalent to the concept of justification itself, because they neglect its prescriptive
meaning.
The third important consequence of the evaluative nature of epistemic justification is this:
Any attempt to naturalize epistemology must be wrong. Although epistemic subjects do have
certain standards by which they evaluate beliefs as justified or unjustified, the meanings of
these terms are not identical to these factual standards. No particular epistemic standard can
possibly be semantically equivalent to the notion of justification, because the latter is primarily used as a device for commending, while standards, however common they may be, just
deliver descriptive features of epistemic judgements. Epistemology is a normative discipline
and to take our or someone’s actual epistemic standards for the meaning of “justified” is to
commit the analogue to the Naturalistic Fallacy in Ethics. So, these metaepistemological con-
1
siderations deliver us with some adequacy conditions for substantial theories of epistemic
justification. Now for the truth predicate.
The truth predicate has to be primarily understood as a prescriptive predicate as well. I try
to establish its evaluative character by arguing for an identification of ideal epistemic justification and truth. My point here is: as “best” is the superlative of “good”, “true” is the superlative of “justified”. Epistemic justification is indeed commonly understood as fallible, but
since justification is a predicate of epistemic commendation, it can only be sincerely used if
justification bears some relation to truth. Usually this is put in the words that justification is
truth conducive. But it’s hard to see how justification could have some relation to truth if truth
was something entirely out of our cognitive scope. Since justification is a gradable notion, it
must at least in principle be reasonable to accept the idea of an ideal form of epistemic justification, which we take as the basis of our evaluative judgements. I call this “infallible justification”. Sometimes certain actual justification procedures just are treated as infallible by us,
because otherwise we couldn’t explain the fact that we actually call some other justification
procedures fallible. No falsification without verification, and, indeed, we (sometimes) call
some of our beliefs true. Since we understand ourselves as epistemic subjects that have no
direct access to the truth (there’s no such thing as divine illumination) we’re always dependent on justification, which would be in these cases treated as infallible.
If truth can be thus identified with a species of epistemic justification, namely an epistemic commendation of a superlative kind, then the characteristics of being a normative and
supervenient predicate must extend to the truth predicate as well. This also results in an important adequacy condition of substantial truth theories:
First: The truth predicate is not (primarily) a predicate in the usual sense of the word, i.e.
we don’t use it (primarily) to describe sentences or relations of sentences and ‘facts’ or something else. This is a blow against almost all old-fashioned theories of truth, mainly against
correspondence theories. This is not to say that truth predications cannot be used to make descriptions sometimes. But in these cases the secondary, descriptive meaning of “true” is at
forehand, which consist in the indication of some epistemic standard. So, when we make descriptive use of “true” in “It is true that snow is white” we say something to the effect, e.g.:
“I’ve seen it with my own eyes that snow is white” or “All people which I take to be reliable
accord in that snow is white” or some different epistemic standard.
Second: But also Deflationist, Semantical and Minimalist theories of truth are inadequate
insofar as a complete list of T-Sentences for a given language is not all there is to say to what
truth is. Analogous to Naturalism in epistemology these positions neglect the prescriptive
element of truth predications – and it is precisely this prescriptive force which is the most
important (in the sense of: most general) feature of the truth predicate. In other words these
theories cannot account for why truth is called a norm and used as such.
Nonetheless, truth predications always have or are at least related to a descriptive content,
since we use truth predications to teach epistemic standards. Thus, if I call this belief and that
belief true, the recipient - as long he is capable of finding out common descriptive properties
of the beliefs I call “true” – will understand that I am trying to commend beliefs of a certain
kind to him (e.g. all beliefs that result from visual experience). If “true” would not have this
prescriptive element of meaning, then this latter operation of teaching and learning epistemic
standards would be impossible. If “true” was solely a descriptive predicate then all what one
would saying by “All beliefs which have the descriptive property X are true” was “All beliefs
which have the descriptive property X have the property X” but not something to the effect
of: “All beliefs which have the descriptive property X should be believed”.
I also defend this account of epistemic justification and truth against several objections
against Non-Cognitivist theories from Austin, Grice, Horwich and the Frege-Geach-Objection
(although I am no Non-Cognitivist in the narrower sense, since I am not claiming that evaluative sentences are not used assertively at all). I also remove qualms against the normative
2
conception of epistemology which are raised by Doxastic Involuntarists (who say that beliefs
are not subject to the “will” and therefore, that normative epistemology is untenable).
3